Effects of Health Responsibility Frames: Testing a Mediation Model of Attributions, Emotions, and Social Support Intentions

Abstract

When news stories cover health and illness, they often address issues of responsibility. These responsibility frames can affect recipients’ responsibility beliefs (i.e. attributions) and thereby indirectly affect emotions and motivation to support people affected by health problems. To date, it is not fully understood how responsibility frames affect social support intentions, and if attributions and emotions mediate this effect. In an online experiment with N = 1,088 German participants, we tested the effects of responsibility frames (individually controllable vs. non-controllable) for type 2 diabetes and depression on social support intentions through responsibility attributions and emotional reactions. Mediation analyses show that responsibility frames indirectly affect social support intentions through social-societal attributions and sympathy. This mediation effect was observed in both depression and type 2 diabetes, despite issue-specific differences in attributions, emotions, and social support intentions. We discuss these findings considering framing effects research and health reporting.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10810730.2023.2232326

Notes

¹ According to attribution theory (Weiner, 2006), this is an indirect effect. Thus, we do not expect a total effect of the responsibility frames on social support intentions without including mediators into the model.

² The online experiment was originally conducted in a 4 (individual responsibility frame/social network responsibility frame/societal responsibility frame/medical frame) × 2 (type 2 diabetes/depression) between-subjects design. To test the attribution path hypotheses, the control group was excluded, while the remaining groups were merged into a two-level factor consisting of controllable (individual) vs. non-controllable (social network, society) responsibility frames (see supplementary Table 2).

³ Excluded from mediation analyses for consistency.

⁴ A univariate ANOVA with post hoc comparisons (Bonferroni corrected) demonstrated that social-societal attributions did not significantly differ between the social and societal responsibility frames (M _{Social Network} = 4.72, SD = 1.16, N = 275, M _Society = 4.70, SD = 1.20, N = 260, p > .05, F(3,1084) = 16.56, η² = .04), but only compared to the individual (M _Individual = 4.22, SD = 1.19, p < .001, N = 286) and control frame (M _Control = 4.19, SD = 1.17, p < .001, N = 267). Thus, the combination of the social and societal frames was deemed appropriate.

⁵ All stimulus articles with highlighted responsibility frames can be requested from the first author or accessed via the following link: https://osf.io/d8zer/?view_only=804eddd843d14919bb6671ccce5c9713

⁶ Means of multiple manipulation check items were in the predicted directions, with ANOVA indicating significant differences between the framing groups. See supplementary Table 3 and supplementary Table 4 for the ANOVA results.

⁷ A principal axis factoring EFA with varimax rotation for depression and diabetes attributions (KMO = .913/.910, Bartlett χ² (df 325) = 5872.02/6396.92, p < .001) showed a different factor structure for diabetes (four factors) and depression (five factors). Since we aimed to test our model for both health issues, we decided to align the attributions with the experimental manipulation (individually controllable vs. social-societal frames) rather than calculating mean indices based on the 4- or 5-factor models. Thus, we combined the factors including family, work, and society items to match the frames: individually controllable attributions versus social-societal attributions.

⁸ Equivalent to A-levels or high school diploma.

⁹ Mediation analyses to test H2 and H3 were conducted with the Hayes Macro (Model 82; Hayes, 2018) in SPSS, which calculates standardized and partially standardized path coefficients using the ordinary least squares (OLS) method for total, direct, and indirect effect. The factor responsibility frame was dummy-coded (1 = non-controllable, i.e., social and societal responsibility frames, 0 = individually controllable responsibility frame). Bootstrapping with 5,000 iterations and heteroskedasticity-consistent standard errors (HC4) were used to determine confidence intervals (CI). Indirect effects are interpreted as significant if 0 is not included CI.

Additional information

Funding

The work was supported by the Deutsche Forschungsgemeinschaft [404881979].